SEMG (Surface EMG) is one of the bio-electric signals that are generated from the muscle. There are many kind of muscle in a human body, and used muscles are different for each body movement. Therefore, any kind of body motions can be detected by analyzing generated SEMG patterns. Our objectives of this study are the development of a method that makes perfect detection of the hand motions possible using SEMG patterns. And applying this developed method to the man-machine interface. In this study, our target machine that is controlled by SEMG is a robot arm that is made for industrial use. The system analyzes user's SEMG and discriminate 8 hand motions. Then the system controls a robot arm using discriminating results. This system was Studied with some subjects, and most of them able to control a robot.